Arthur, you are making it harder than it has to be, I think. In the
describe of the
mib variable, check the datatype. You will see things like guage, integer,
etc.
Take a look at /usr/OV/conf/mib.coerce, and find an explanation of how
values
are massaged for you by netview when they are collected. Most of the math
has
already been done for you.
Here's an old append..
===================================================================
To: Ron Jakeman/South Africa/ibm
cc:
From: Leslie Clark/Southfield/IBM@IBMUS
Subject: Re: Collecting Statistics
Hello, Ron. Netview handles the polling interval for you, since you could
change that any time you
wanted, and actually polling intervals may vary due to outages or traffic.
The entries I defined are
only to say what to collect. In the Data Collection & Thresholding dialog
you specify how often and
from whom and which interfaces. Is that what you mean? There is a little
bit of explanation in the
file /usr/OV/conf/mib.coerce. Netview calculates the value at t2 minus the
value at t1 divided by
the interval and also handles wrapped values.
Cordially,
Leslie A. Clark
IBM Global Services - Systems Mgmt & Networking
(248) 552-4968 Voicemail, Fax, Pager
---------------------- Forwarded by Leslie Clark/Southfield/IBM on 03-25-99
06:53 AM ---------------------------
Ron Jakeman@IBMZA
03-25-99 01:39 AM
To: Leslie Clark/Southfield/IBM@IBMUS
cc:
From: Ron Jakeman/South Africa/IBM@IBMZA
Subject: Re: Collecting Statistics
Some comments on measuring utilization. The utilization expressions you
gave report utilization over the polled interval. Isn't it important to
always state the polling interval? Shouldn't you normalize the utilization
by dividing it by the polling interval in seconds - this would avoid
utilization values of over 100% which are often reported if the interval is
not large. Does this make sense?
---------------------- Forwarded by Ron Jakeman/South Africa/IBM on
03/25/99 08:31 AM ---------------------------
Please respond to Discussion of IBM NetView and POLYCENTER Manager on
NetView <NV-L@UCSBVM.UCSB.EDU>
To: NV-L@UCSBVM.UCSB.EDU
cc: (bcc: Ron Jakeman/South Africa/IBM)
Subject: Re: Collecting Statistics
Expression in /usr/OV/conf/mibExpr.conf (from my notebook...)
For lans, measure both send and receive, since data only flows in one
direction at a time (half-duplex):
LANIf%Util \
"% interface utilization from \
((ifInOctets + ifOutOctets) * 8 * 100) / ifSpeed" \
.1.3.6.1.2.1.2.2.1.10. \
.1.3.6.1.2.1.2.2.1.16. + \
8 * 100 * \
.1.3.6.1.2.1.2.2.1.5. /
For WANs, measure separately, since data flows both ways at once
(full-duplex):
WANIf%RecvUtil -
"% interface utilization from \
(ifInOctets * 8 * 100) / ifSpeed" \
.1.3.6.1.2.1.2.2.1.10. \
8 * 100 * \
.1.3.6.1.2.1.2.2.1.5. /
WANIf%SendUtil -
"% interface utilization from \
(ifOutOctets * 8 * 100) / ifSpeed" \
.1.3.6.1.2.1.2.2.1.16. \
8 * 100 * \
.1.3.6.1.2.1.2.2.1.5. /
My notes on what to use when for WANs:
A 56kb line can send 56kb and receive 56kb at the same time. So in one
sense its capacity is 112kb, but not really, since neither can use more
than 56kb. Two methods:
1) Take the highest: util = max(wansend,wanrecv)
This tells you how close you are to the limit, regardless of the
direction that is being a pig.
2) Take the average or double the speed:
util=8*(inoctets+outoctets)/(ifspeed*2)*100
This gives effective util of entire link, but is deceptive about how
close you are to capacity. Eg 99% send and 21% receive would show as
60% util, but you are really maxxed on the send side.
Cordially,
Leslie A. Clark
IBM Global Services - Systems Mgmt & Networking
(248) 552-4968 Voicemail, Fax, Pager
---------------------- Forwarded by Leslie Clark/Southfield/IBM on 03-23-99
12:38 AM ---------------------------
Frantsen Christian <cf@INTERNOC.SE> on 03-22-99 05:09:08 AM
Please respond to Discussion of IBM NetView and POLYCENTER Manager on
NetView <NV-L@UCSBVM.UCSB.EDU>
To: NV-L@UCSBVM.UCSB.EDU
cc: (bcc: Leslie Clark/Southfield/IBM)
Subject: Collecting Statistics
Hi!
Im about to start doing some utilization calcultions on several different
connections (2mbit Radio LAN, 10/100Mbit Ethernet, 100Mbit FDDI) and was
wondering if someone has made some good ones in mibExpr.conf they would
like
to share :)
I'm rather new at this SNMP-stuff so it would be nice to get a chance to
have a look at how ppl are doing this, it would help me understand it
better.
Thanx in advance.
-----------------------------------------
Christian Frantsen
Internoc Scandinavia AB
Tel: +46-36-194843
Fax: +46-36-194651
Leslie A. Clark
IBM Global Services - Systems Mgmt & Networking
(248) 552-4968 Voicemail, Fax, Pager
---------------------- Forwarded by Leslie Clark/Southfield/IBM on
01/28/2000 07:37 AM ---------------------------
"Boulieris, Arthur" <Arthur.Boulieris@NZ.UNISYS.COM>@UCSBVM.UCSB.EDU> on
01/27/2000 08:56:27 PM
Please respond to Discussion of IBM NetView and POLYCENTER Manager on
NetView <NV-L@UCSBVM.UCSB.EDU>
Sent by: Discussion of IBM NetView and POLYCENTER Manager on NetView
<NV-L@UCSBVM.UCSB.EDU>
To: NV-L@UCSBVM.UCSB.EDU
cc:
Subject: ifinoctet data format
Hello
Im running Netview 5.1.2 on solaris 2.6.
I use it for collecting ifinoctets and ifoutoctets and store the data in
oracle via RIM.
I now have a month data and am working on some end of month reports and am
puzzled by the numbers.
My calculation is add ifinoctet + ifoutoctet to get the total bandwidth
utilsation.This figure is typically around 4000.00 to 8000.00.Does anyone
know what the two decimal places are for?
Also on say a value of 8000 octets per 10 minutes(my collection interval)
the bits / sec works out to be 8000 x 8 to get total bits per 10
minutes.Then divide by 600 to get bits/sec.
This gives a value of 8000 x 8 / 600 or 106.6 bits/sec.This all seems
wrong
I expect it to be around 20 kb/s.
Am I misinterpreting the data?
I also noticed that some of the collections have a value to 8 decimal
places.Why?
If anyone out there is reporting on similar stats I would appreciate your
comments.
Regards Arthur
Arthur Boulieris
Implementation & Support
Systems Management
<<...>>
|